Cheap Claude vs cheap GPT
Claude Haiku 5.5 vs GPT-6 Luna
Claude Haiku 5.5 (Anthropic) and GPT-6 Luna (OpenAI) compared on the facts each vendor publishes. Price does not decide it: both list $0.1 in and $0.5 out per 1M tokens. For very long inputs, GPT-6 Luna lists the larger context window: 1.05M tokens against 1M. We have not run either model: every value below is Reported, with a link to its source.
Reported Updated · Markdown · All model comparisons
Anthropic · generally available · API only · released Oct 7, 2026
- $ in / 1M tokens
- $0.1
- $ out / 1M tokens
- $0.5
- Context
- 1M
- Max output
- —
OpenAI · generally available · API only · released Sep 22, 2026
- $ in / 1M tokens
- $0.1
- $ out / 1M tokens
- $0.5
- Context
- 1.05M
- Max output
- 128K
How to choose
- Price does not decide it: both list $0.1 in and $0.5 out per 1M tokens.
- For very long inputs, GPT-6 Luna lists the larger context window: 1.05M tokens against 1M.
- Claude Haiku 5.5 is newer, by 15 days.
Specs side by side
Each value is read from the vendor page linked beside it. “Not published” means we found no vendor figure, not that the feature is missing.
| Spec | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Price (as listed) | $0.1 in / $0.5 out per 1M tokenssource | $0.1 in / $0.5 out per 1M tokenssource |
| Context window | 1,000,000 tokens (1M)source | 1,050,000 tokens (1.05M)source |
| Max output | Not published | 128,000 tokenssource |
| Input | text, imagesource | text, imagesource |
| Output | textsource | textsource |
| Licence | Proprietary (API)source | Proprietary (API)source |
| Released | Oct 7, 2026source | Sep 22, 2026source |
| Status | generally available | generally available |
| API id | claude-haiku-5-5source | gpt-6-lunasource |
Benchmark snapshots
From the boards named below, not our tests. Boards use different tasks and settings, so compare within a row, never across rows.
| Board | Claude Haiku 5.5 | GPT-6 Luna | Type, date |
|---|---|---|---|
| Terminal-Bench 4.0 official leaderboard | not listed | 16.4% | Independent, Oct 26, 2026 |
| SWE-bench Pro (Public), as reported by the vendors | 64.8% | not listed | Vendor-reported, Oct 26, 2026 |
More context: best LLM for coding.
FAQ
Which is cheaper, Claude Haiku 5.5 or GPT-6 Luna?
Claude Haiku 5.5: $0.1 in / $0.5 out per 1M tokens. GPT-6 Luna: $0.1 in / $0.5 out per 1M tokens. Both prices are the vendors' own listed prices on the date each model was last verified.
Which has the bigger context window?
GPT-6 Luna, with 1,050,000 tokens against 1,000,000.
Which is better for coding?
We have not tested either. Sourced snapshots: Terminal-Bench 4.0 official leaderboard (independent, Oct 26, 2026): GPT-6 Luna (Codex, max) 16.4%; SWE-bench Pro (Public), as reported by the vendors (vendor-reported, Oct 26, 2026): Claude Haiku 5.5 (max) 64.8%. Scores from different boards are not comparable with each other.
Are Claude Haiku 5.5 and GPT-6 Luna open source?
Claude Haiku 5.5: Proprietary (API), no downloadable weights. GPT-6 Luna: Proprietary (API), no downloadable weights.
Did you test Claude Haiku 5.5 and GPT-6 Luna?
No. This page is Reported: every fact is read from the vendor page linked beside it, and benchmark numbers come from the named boards.
Related
The Weekly Verdict · free
5 new-model use cases. Every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.